Why trust these numbers.

Every figure on this site can be recomputed from published data — including the ones that argue against us. This page is the index of the receipts.

We publish our own model losing. Across 2,671 graded walk-forward NFL games (2,608 decided, 63 pushes) our prediction model hit 49.5% of decided games against the spread; the market hit 50.9%; breakeven is 52.4%. Nobody clears it. That measurement is why Sooth sells price transparency instead of picks — and why you can trust the rest of this site: numbers that flatter us and numbers that don't get published the same way.

The methodMethodology

The full model spec, the walk-forward backtest, the calibration table, the leakage controls — the file we'd hand a regulator or a journalist.

Read it →
The receiptsPublic ledger

Board snapshots hash-committed before kickoff, revealed and graded after. Nothing can be quietly edited later.

See the ledger →
Check usVerify independently

Recompute our Merkle root yourself with a standard-library script. If a number can't be regenerated from published data, treat it as an error and tell us.

Run the check →
The machineEngine room

27 seasons of data, the honest backtest curve, and the measured +2.71% price edge — what the archive actually buys us.

Open the engine →
The archiveGame log

Every NFL game since 1999 with its closing spread, total, and moneylines — the raw history behind every claim.

Browse 7,548 games →
The fine printDisclaimers

No wagers, no funds held, no performance claims, no affiliation with any league or book. Written plainly.

Read them →

Analysis only — no wagers taken, no picks sold. 21+, where lawful. If gambling is causing harm, call 1-800-522-4700, 24/7.